A5020615530- Recommender Systems and Techniques
- Advanced Bandit Algorithms Research
- Nuclear physics research studies
- Image Retrieval and Classification Techniques
- Advanced Graph Neural Networks
- Video Analysis and Summarization
- Advanced NMR Techniques and Applications
- Topic Modeling
- Web Data Mining and Analysis
- Multimodal Machine Learning Applications
- Protein Structure and Dynamics
- Visual Attention and Saliency Detection
- Machine Learning in Healthcare
- Machine Learning in Bioinformatics
- Data Stream Mining Techniques
- Fault Detection and Control Systems
- Consumer Market Behavior and Pricing
- Customer churn and segmentation
- Information Retrieval and Search Behavior
- Advanced Text Analysis Techniques
- Stock Market Forecasting Methods
- Human Pose and Action Recognition
- Human Mobility and Location-Based Analysis
- Image and Video Quality Assessment
- Natural Language Processing Techniques
Kuaishou (China)
2024-2025
Dartmouth Hospital
2020
In recommender systems, reinforcement learning solutions have shown promising results in optimizing the interaction sequence between users and system over long-term performance. For practical reasons, policy's actions are typically designed as recommending a list of items to handle users' frequent continuous browsing requests more efficiently. this list-wise recommendation scenario, user state is updated upon every request corresponding MDP formulation. However, request-level formulation...
Recent advances in recommender systems have shown that user-system interaction essentially formulates long-term optimization problems, and online reinforcement learning can be adopted to improve recommendation performance. The general solution framework incorporates a value function estimates the user's expected cumulative rewards future guides training of policy. To avoid local maxima, policy may explore potential high-quality actions during inference increase chance finding better rewards....
In recent years, graph contrastive learning (GCL) has received increasing attention in recommender systems due to its effectiveness reducing bias caused by data sparsity. However, most existing GCL models rely on heuristic approaches and usually assume entity independence when constructing views. We argue that these methods struggle strike a balance between semantic invariance view hardness across the dynamic training process, both of which are critical factors learning. To address above...
Natural language explainable recommendation has become a promising direction to facilitate more efficient and informed user decisions. Previous models mostly focus on how enhance the explanation accuracy. However, robustness problem been largely ignored, which requires explanations generated for similar user-item pairs should not be too much different. Different from traditional classification problems, improving of natural languages two unique characteristics: (1) token importances, that...
Short video has witnessed rapid growth in the past few years multimedia platforms. To ensure freshness of videos, platforms receive a large number user-uploaded videos every day, making collaborative filtering-based recommender methods suffer from item cold-start problem (e.g., new-coming are difficult to compete with existing videos). Consequently, increasing efforts tackle issue content perspective, focusing on modeling multi-modal preferences users, fair way and videos. However, recent...
In this work, the beta-decay halflives problem is dealt as a nonlinear optimization problem, which resolved in statistical framework of Machine Learning (LM). Continuing past similar approaches, we have constructed sophisticated Artificial Neural Networks (ANNs) and Support Vector Regression Machines (SVMs) for each class with even-odd character Z N to global model systematics nuclei that decay 100% by beta-minus-mode their ground states. The arising large-scale lifetime calculations...
Recommender systems have emerged as an indispensable mean to meet personalized interests of users and alleviate information overload. Despite the great success, accuracy-oriented recommendation models are creating cocoons, i.e., it is becoming increasingly difficult for see other items they might be interested in. Although recent studies start paying attention enhancing diversity, based on point embedding fail describe range user preferences item features well, which essential diversified...
Recommender systems aim to fulfill the user's daily demands. While most existing research focuses on maximizing engagement with system, it has recently been pointed out that how frequently users come back for service also reflects quality and stability of recommendations. However, optimizing this user retention behavior is non-trivial poses several challenges including intractable leave-and-return activities, sparse delayed signal, uncertain relations between users' their immediate feedback...
Rich user behavior data has been proven to be of great value for recommendation systems. Modeling lifelong in the retrieval stage explore long-term preference and obtain comprehensive results is crucial. Existing modeling methods cannot applied because they extract target-relevant items through coupling between target item. Moreover, current fail precisely capture interests when length sequence increases further. That leads a gap ability models model data. In this paper, we propose concept...
E-commerce is increasingly multimedia-enriched, with products exhibited in a broad-domain manner as images, short videos, or live stream promotions. A unified and vectorized cross-domain production representation essential. Due to large intra-product variance high inter-product similarity the scenario, visual-only inadequate. While Automatic Speech Recognition (ASR) text derived from live-stream videos readily accessible, how de-noise excessively noisy for multimodal learning mostly...
In this work, the beta-decay halflives problem is dealt as a nonlinear optimiza- tion problem, which resolved in statistical framework of Machine Learning (LM). Continuing past similar approaches, we have constructed sophisticated Artificial Neural Networks (ANNs) and Support Vector Regression Machines (SV Ms) for each class with even-odd character Z N to global model systemat- ics nuclei that decay 100% by β−-mode their ground states. The arising large-scale lifetime calculations generated...